Bug 1240261

Summary:	Libvirt's host UUID is the same for all Nova-compute nodes
Product:	Red Hat OpenStack	Reporter:	Yogev Rabl <yrabl>
Component:	openstack-tripleo	Assignee:	James Slagle <jslagle>
Status:	CLOSED NOTABUG	QA Contact:	nlevinki <nlevinki>
Severity:	urgent	Docs Contact:
Priority:	high
Version:	Director	CC:	calfonso, mburns, rhel-osp-director-maint, ukalifon, yeylon, yrabl
Target Milestone:	z2	Keywords:	Triaged, ZStream
Target Release:	Director
Hardware:	Unspecified
OS:	Unspecified
Whiteboard:
Fixed In Version:		Doc Type:	Bug Fix
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2015-08-24 06:53:23 UTC	Type:	Bug
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:

Description Yogev Rabl 2015-07-06 11:37:21 UTC

Description of problem:
The host_uuid parameter is commented out in Libvirt's configuration file -  /etc/libvirt/libvirtd.conf. The result is all of the Nova compute nodes has the same unique ID. 

A similar bug was closed Bug 1192123. 

Version-Release number of selected component (if applicable):
python-rdomanager-oscplugin-0.0.8-13.el7ost.noarch

How reproducible:
100%

Steps to Reproduce:
1. Install compute nodes 

Actual results:
All of the Nova Compute nodes has the same unique ID. 
That will cause issues with migrations of instances between hosts. 

Expected results:
Each node have a different unique ID

Additional info:

Comment 3 Mike Burns 2015-07-06 15:31:13 UTC

*** Bug 1240322 has been marked as a duplicate of this bug. ***

Comment 4 chris alfonso 2015-07-06 16:20:30 UTC

This is only on virt, correct?

Comment 5 Udi Kalifon 2015-07-06 16:48:58 UTC

No it's on bare metals also.

Comment 6 James Slagle 2015-07-06 17:41:44 UTC

host_uuid is commented out in /etc/libvirt/libvirtd.conf by default. I don't follow why we need to change that.

The comments above the config option say:
###################################################################
# UUID of the host:
# Provide the UUID of the host here in case the command
# 'dmidecode -s system-uuid' does not provide a valid uuid. In case
# 'dmidecode' does not provide a valid UUID and none is provided here, a
# temporary UUID will be generated.
# Keep the format of the example UUID below. UUID must not have all digits
# be the same.

So, the output of dmidecode -s system-uuid will be used by default. And you're saying that you're getting the same value from this command on 2 different machines?

Can you confirm that when you run this command on 2 different deployed compute nodes, you get the same value? Don't use a cloned VM of a compute node to confirm this.

You can also ask virsh directly for the uuid value:

# virsh sysinfo

The uuid value is shown in the output.

Comment 10 Yogev Rabl 2015-07-09 12:13:38 UTC

This problem though persists in former installers doens't reproduce all the times in the director. Perhaps the Directors different installation method fixed the problem

Comment 11 chris alfonso 2015-08-21 16:51:08 UTC

Can you confirm this is no longer an issue (installed with OSP-director)

Comment 12 Yogev Rabl 2015-08-24 06:53:23 UTC

After some rechecking I've found the root cause of this issue, libvirt's system UUID is been set by the output of the command 'dmidecode -s system-uuid'. In my case I have two servers that are in the same chassis, both have the same UUID.
From what I can say, this is not a bug in the installer, but a bug in libvirt.